Orchestrates multiple Codex agents for parallel work while preventing file conflicts through upfront target analysis. You break down tasks, identify which files each touches, then merge same-file edits into one agent or sequence them across waves if there's partial overlap. Auto-selects between Codex experimental sub-agents, CLI backend, or OpenCode subagents depending on what's available. Includes JSONL progress monitoring and cross-project support via --add-dir. The core insight is simple: headless agents can't negotiate locks, so the team lead acts as lock manager before spawning. Works best for 2+ well-scoped tasks that don't need tight shared state coordination. Falls back to /swarm for dependency-heavy orchestration.
npx skills add https://github.com/boshu2/agentops --skill codex-team